No abstract

This study presented an insulation state monitoring method for large generator based on radio frequency (RF) technique. As an on-line condition monitor and the precondition of condition-based maintenance (CBM), the RF monitor used the high frequency current mutual inductor to detect the partial discharge signal from neutral wire of stator windings. According to the magnitude of indicative value of RF monitor, a five phase model was also proposed to manage the generator's running better. The practices show that the proposed method is effective.

Through the analysis of the electromagnetic properties for the rotor coil of turbine generator, we obtained the main magnetic field variation characteristics for the rotor winding and put forward the short circuit fault between rotor inter-turns causing the potential difference and circulation between the parallel branches of generation stator winding spectral method was applied on the spectrum analysis for the current signal of generator stator winding and spectral feature vector was treated as learning sample. Through training, the RBF neural network can reflect the mapping relations between spectral features and fault types so as to achieve the objective of fault diagnosis. The practical application showed that the integration of spectral analysis method and RBF neural network can effectively improve the diagnostic accuracy and efficiency.